Download Uranium enrichment technologies

FFI-rapport 2008/02376
Uranium enrichment technologies
Siv Gundrosen Aalbergsjø
Norwegian Defence Research Establishment (FFI)
6 July 2009
FFI-rapport 2008/02376
Oppdrag 3423
P: ISBN 978-82-464- 1688-5
E: ISBN 978-82-464- 1689-2
Keywords
Uran
Uran - Anriking
Approved by
Steinar Høibråten
Research Manager
Monica Endregard
Project Manager
Jan Ivar Botnan
Director
2
FFI-rapport 2008/02376
Summary
The enrichment of uranium has taken place since World War II, initiated by the quest for the
atomic bomb to win the war. To enrich uranium means to separate the two main isotopes that
make up natural uranium (U-235 and U-238). The details of the possible enrichment processes are
generally classified because highly enriched uranium may be used for weapons production, but
the principles are well known.
The gaseous diffusion method developed during the war constitutes the first generation of
uranium enrichment techniques and has been used both for enrichment to weapons grade uranium
and also for production of low-enriched uranium for nuclear power plants. This technique has
high energy consumption and takes up a lot of space and is therefore being phased out in favour
of gas centrifuges. Gas centrifuges take up less space and use only about one tenth of the energy
of the gaseous diffusion method. Gas centrifuges are harder to build technically, but once they are
built they can more easily be used to produce highly enriched uranium, as well as low-enriched
uranium, for weapons or other purposes. The gas centrifuge technology is spread around the
world and has even been for sale on the black market. Laser isotope separation methods are about
to become the third generation of commercial scale uranium enrichment technologies. These
methods are researched by many countries, and have been developed to a great extent. The energy
consumption of these processes is about the same as for gas centrifuges, but the facilities require
much less space than the centrifuge plants. These are high-precision technologies, and making the
lasers and other necessary equipment has posed significant technical challenges. It appears
difficult to expand from laboratory scale to large scale facilities, and most of the research has
been terminated. The only still active laser process is the SILEX process which is about to go
commercial in a few years.
Another technique which has been used on a large scale is the aerodynamic technique used in
South Africa. This technique was used both for production of low-enriched and highly enriched
uranium, but it is not regarded as commercially competitive because of its high energy
consumption. Also the mass spectrometer method was used in a relatively large scale in the
United States during World War II, but this process is too slow and consumes too much energy to
be commercially competitive as well. The mass spectrometer method is relatively easy to
perform, and could therefore be used in a non-commercial setting as a slow but reliable option for
enrichment. This could also be said about the thermal diffusion method which was used to
produce feed for the mass spectrometer method in the United States during World War II. The
thermal diffusion method could not be used for high-level enrichment. Plasma processes have not
been applied on a large scale yet. The chemical enrichment processes developed in France and
Japan have been utilised in pilot plants, but not yet in full-scale facilities.
FFI-rapport 2008/02376
3
Norwegian summary
Urananrikning fant først sted i forbindelse med utviklingen av kjernevåpen under andre
verdenskrig. Anrikning av uran innebærer at en skjelner mellom de to viktigste isotopene i
naturlig uran (U-235 og U-238). Detaljene i de mulige anrikningsprosessene er generelt gradert
fordi høyanriket uran kan brukes i kjernevåpen, men prinsippene er velkjente.
Gassdiffusjonsmetoden som ble utviklet under andre verdenskrig, utgjør første generasjon
urananrikningsteknologi og har vært anvendt både til å lage våpenmaterialer og til å lage
lavanriket uran til kjernekraftverk. Denne metoden krever mye energi og store anlegg og blir
derfor faset ut til fordel for gassentrifuger. Gassentrifuger er mindre plasskrevende og krever
bare om lag en tidel så mye energi som gassdiffusjonsmetoden. Sentrifugene er teknisk krevende
å lage, men når de først er på plass, er de enklere å benytte enn diffusjonsutstyr til produksjon av
både lavanriket og høyanriket uran. Gassentrifugeteknologien er i bruk mange steder i verden og
har også vært tilgjengelig på det svarte markedet. Laserisotopseparasjonsmetoder er i ferd med å
bli tredje generasjon av kommersielle urananrikningsteknologier. Det er blitt forsket på disse
metodene i en rekke land, og en god del utviklingsarbeid har funnet sted. Energibehovet er
omtrent som for gassentrifugemetoden, men anleggene er vesentlig mindre plasskrevende.
Lasermetodene krever avansert teknologi med høy presisjon, og det har vist seg svært utfordrende
å lage både laserne og annet nødvendig utstyr. Det har vært vanskelig å gå fra laboratorieskala til
industriell skala, og mye av forskningen har blitt avsluttet. Den eneste aktive lasermetoden er den
såkalte SILEX-prosessen som forventes å bli kommersielt tilgjengelig i løpet av noen få år.
En annen metode som har vært anvendt i stor skala, er aerodynamisk urananrikning, som SørAfrika benyttet seg av. Der ble metoden brukt både til lavanriket og høyanriket uran, men den
anses ikke kommersielt konkurransedyktig fordi den er svært energikrevende. Under andre
verdenskrig ble også massespektrometermetoden anvendt i relativt stor skala i USA, men denne
metoden er for langsom og for energikrevende til å være kommersielt konkurransedyktig.
Metoden er relativt enkel og kan derfor tenkes anvendt som en langsom, men pålitelig, mulighet i
ikke-kommersielle situasjoner. Det samme kan sies om metoden med termisk diffusjon som ble
brukt til å produsere utgangsmaterialet for de amerikanske massespektrometrene under andre
verdenskrig. Termisk diffusjon kan ikke anvendes til å lage høyanriket uran. Plasmaprosesser
har ennå ikke blitt anvendt i stor skala. Kjemiske anrikningsmetoder har blitt utviklet i Frankrike
og Japan. De har vært demonstrert i pilotanlegg, men ikke i industriell skala.
4
FFI-rapport 2008/02376
Contents
1
Introduction
7
2
Uranium enrichment
7
2.1
How enrichment processes work
8
3
Gaseous diffusion
10
3.1
The main principle
10
3.2
The enrichment plant
10
3.2.1
The separation stage
11
3.2.2
The diffusion barrier
12
3.2.3
Energy consumption
13
3.3
When and where has the technique been used?
13
4
The gas centrifuge method
13
4.1
The main principle
13
4.2
Technical challenges
14
4.2.1
Building the centrifuge
14
4.2.2
Energy consumption and space requirements
17
4.2.3
Feasibility for production of HEU
18
4.3
When and where has the technique been used?
18
5
Thermal diffusion
19
5.1
The main principle
19
5.2
When and where has the technique been used?
20
6
Electromagnetic processes
20
6.1
Mass spectrometer: EMIS (calutron)
21
6.1.1
The main principle
21
6.1.2
Equipment and efficiency
21
6.1.3
When and where has the technique been used?
22
6.2
Ion cyclotron resonance
23
6.2.1
The main principle
23
6.2.2
How it is done
23
6.2.3
When and where has the technique been used?
24
6.3
The plasma centrifuge method
24
7
Aerodynamic processes
25
7.1
The separation nozzle process
26
7.2
The vortex tube process
27
FFI-rapport 2008/02376
5
8
Laser isotope separation
27
8.1
Atomic vapour laser isotope separation (AVLIS)
28
8.1.1
The main principle
28
8.1.2
Advantages and problems with the AVLIS process
29
8.2
Molecular laser isotope separation (MLIS)
30
8.2.1
The main principle
30
8.2.2
Potential for HEU production
31
8.2.3
Separation of isotopes by laser excitation (SILEX)
31
8.3
Laser assisted processes
32
8.4
When and where have the LIS techniques been used?
32
9
Chemical separation methods
34
9.1
The French process (CHEMEX)
35
9.2
The Japanese process (ACEP)
35
9.3
A third process
36
9.4
When and where has the technique been used?
37
10
Brief summary and comments
37
References
37
Appendix A Some relevant physics and chemistry
41
A.1
Isotopes
41
A.2
Nuclear fission and chain reaction
41
A.3
Uranium hexafluoride
42
A.4
Statistical description of the molecules in a gas
43
A.5
Quantum mechanics
43
A.6
Oxidation states
44
Appendix B Lasers
45
6
FFI-rapport 2008/02376
1
Introduction
Uranium enrichment is the key to making uranium-based nuclear weapons. It is also a process
that is important in the nuclear power industry. In the race for the atomic bomb during World
War II, the United Stated pursued several methods for enriching uranium all at once. Also
Germany and the Soviet Union worked on developing enrichment methods at this time. Some of
these methods proved inefficient and have been discarded. Others, where the most prominent
examples are the gaseous diffusion method and the gas centrifuge method, have been further
developed and made into commercial processes used today for enriching uranium for commercial
nuclear power plants. The enrichment of uranium is considered the most technically and
industrially demanding part of any nuclear program (either civilian or military).
Today most enrichment plants produce low-enriched uranium, but even now almost all
enrichment plants can be modified to produce highly enriched uranium.[1] This report gives an
overview of the most common enrichment technologies and addresses some of the technical and
economical difficulties associated with them as well as when and where these technologies have
been in use. All of these techniques have been researched by more than one country, but not all of
them have made it to actual production. Several other techniques exist either on the drawing
board or have been tried to some extent, but this report covers only the techniques considered
most promising for future use.
To understand the techniques discussed in this report, some basic knowledge of physics and
chemistry is needed. Some of this information is provided in Appendix A.
2
Uranium enrichment
Like all elements, uranium consists of different isotopes (see Section A.1); the two that are the
most abundant and relevant, are known as U-235 and U-238. Natural uranium consists of
approximately 99.3 % U-238 and 0.7 % U-235. When uranium is to be used as fuel for a power
plant or in a bomb, it is the less abundant isotope U-235 that is needed because it is this isotope
that is most likely to fission (see Section A.2) when it absorbs one neutron and releases several
neutrons in each fission process. U-235 is able to sustain a chain reaction if enough of the isotope
is present. The challenge is to find a way to extract this isotope from the natural uranium; this
process is called enrichment. There are several ways of doing this, and some of the most
important techniques are discussed in this report. It is customary to speak of three different levels
of enrichment. Low-enriched uranium (LEU) contains less than 20 % U-235; the fuel used in
power plants is usually enriched to about 3-5 %. Highly enriched uranium (HEU) contains more
than 20 % U-235; uranium enriched to this level is used in some specialised reactors. If the
uranium is enriched beyond about 90 %, it may also be referred to as weapons grade uranium.
FFI-rapport 2008/02376
7
2.1
How enrichment processes work
The different methods for uranium enrichment all have in common that they make use of the
different properties of the isotopes (either physical or chemical) to physically separate U-235
from U-238. No enrichment technology is perfect so the products will never be 100 % U-238 in
one place and 100 % U-235 in another; there will always be two new mixtures, one which is
enriched in U-235, and one which is correspondingly depleted in U-235. Because of its physical
properties, depleted uranium (DU) is commonly used in conventional armour-piercing
munitions, 1 but it is in fact a waste product from the nuclear industry.
The physical property that most enrichment technologies make use of, either directly or
indirectly, is the mass difference between the isotopes. Heavier particles move differently from
lighter particles when experiencing the same external influence, and even though the mass
difference between the two isotopes of uranium is very small, it is possible to exploit this to
separate them from each other. In the gas centrifuge method, gas molecules containing U-238
will concentrate closer to the centrifuge wall than the molecules containing U-235. The gaseous
diffusion method exploits the fact that heavier molecules with a certain energy move slower than
lighter molecules with the same energy and therefore will diffuse more slowly trough a semipermeable membrane. The mass spectrograph separates the isotopes because particles with a
different charge-to-mass ratio will move differently in a magnetic field. Even chemical properties
of a molecule may be affected by the mass of the atoms it consists of.
Much of the information below is from Reference [2].
The separation elements in an enrichment facility and also the full facility may be described in a
general way. In the following discussion some common factors of the various enrichment
facilities will be described, and the elements may be considered as black boxes (it does not matter
which enrichment technique is used). The feed, the form of natural uranium or uranium enriched
to a lower degree than what is desired, is entering at one end, and the product (the enriched
uranium) and the tails or waste (the depleted uranium) emerge at the other end. This terminology
is also used for the entire facility. The separation factor of an element is the ratio of the relative
amount of U-235 in the product to the relative amount in the tails, and is therefore a measure of
the efficiency. Separation factors are just above 1 for each step in the most common enrichment
methods such as gaseous diffusion and gas centrifuges.
Enrichment methods with low separation factors have been made effective by organizing the
elements into stages and cascades. A stage is comprised of one or more elements that are
organized in parallel; this means that they all receive their input from the same source and
contribute to the same output product. A cascade consists of a number of stages organized in a
series; this means that the product of one stage is used as feed for the next and that the waste may
be returned to the previous stage or earlier (known as reflux). The cascade is called ideal if the
1
Uranium has a very high density. Furthermore, it is pyroforic, that is, it ignites and burns as it penetrates
the armour.
8
FFI-rapport 2008/02376
arrangement is such that streams of uranium of different enrichments are never mixed. This
minimizes the ratio between energy input and product gained from the cascade. Ideal cascades
also have the shortest equilibrium time (described below) for a given product enrichment. The
downside of an ideal cascade is that all the stages must be of different size and use a different
amount of power, so construction of an ideal cascade is not economical. In practice a compromise
between what is feasible to construct and idealization is used.
It is also possible to use batch recycling to enhance the level of enrichment of the uranium, this
means that enriched material is run through the enrichment facility one or more times extra to
increase the enrichment level. Which properties of the enrichment facility make this a good or a
bad idea will be mentioned below. Yet another way of increasing the enrichment level of the
product is to use stretching [1]. When stretching is used there is a large reflux and the product is
extracted at a very low rate. This gives a high percentage of U-235 in the product and a very low
percentage of U-235 in the tails at the expense of efficiency and production rates. Batch recycling
or stretching are methods which may be used to enrich to a higher degree than originally intended
for the enrichment facility. If a country already has a plant for enriching uranium for use in
nuclear power reactors and subsequently decides to produce HEU, then these methods could be
used instead of rebuilding the plant or building a new one.
Some factors that are general and therefore suitable for comparing different enrichment
technologies are the equilibrium time, the inventory, and the kilogram separative work units
(SWU 2 ). When a cascade is started, all the stages are filled with material of the same isotopic
composition; therefore it takes a while before the product reaches the desired enrichment. Until
this point is reached, the cascade is operating in total reflux (no product or tails are extracted).
The equilibrium time is a measure of the time it takes from the initial start-up until any product
can be extracted. To use batch recycling to gain a higher level of enrichment, the whole cascade
must be emptied, refilled and started up again, so the equilibrium time is an important factor in
determining whether this is suitable. The inventory (the amount of uranium it takes to fill the
cascade) is of course also important in this discussion. To reach weapons grade uranium from
LEU by batch recycling, the batch will have to be recycled a few times. Batch recycling can be
wasteful in uranium because some material will be lost in each recycling.
The kilogram separative work units that a facility can produce per year depend on the feed flow
(the throughput) and the separation factor in each stage. The unit is useful for comparing different
technologies when there are many stages and a small separation factor per stage, for example in
the case of centrifuges and gaseous diffusion. If the percentage of U-235 in the feed, the product
and the tails are all set, then the number of SWUs produced by the facility is directly related to
the amounts of product, feed and tails. For instance, in order to produce 25 kg of 93 % U-235,
5000 SWU and 5050 kg of natural uranium feed is needed if the tails contain 0.3 % U-235 [1]. In
commercial processes, the costs are generally directly related to the number of SWUs, and so this
has come to be a commonly used parameter for describing enrichment facilities.
2
The kilogram separative work unit is abbreviated kg SWU or just SWU [1].
FFI-rapport 2008/02376
9
When enriching uranium, one important aspect is criticality, the amount of enriched uranium may
become large enough to sustain a nuclear chain reaction (see Section A.2). For low-level
enrichment, criticality is not a problem because the concentration of U-235 will not be high
enough. When producing HEU, the issue is present; it is important not to reach critical mass or
critical density. If the inventory is large, this problem is more prominent than if the inventory is
small, either way pressure must be controlled as to not reach critical density.
The most common uranium compound utilised in enrichment facilities is uranium hexafluoride
(UF6) gas (see Section A.3). This gas is highly corrosive; therefore all the parts of machines that
come into contact with the gas must be corrosion resistant. Uranium hexafluoride is used in the
gaseous diffusion method, the gas centrifuge method, thermal diffusion, aerodynamic processes
and molecular laser isotope separation. Uranium hexafluoride gas is a good molecule for uranium
isotope separation because the molecule is simple, and the mass difference between the uranium
atoms is the only source of mass difference between the molecules (fluoride has only one stable
isotope). Uranium hexafluoride is also preferred because of the extensive experiences
accumulated by working with it (it has been used for the enrichment of uranium since the early
1940s). Other techniques employ uranium in the form of uranium tetrachloride (UCl4) or in its
pure metallic form.
3
Gaseous diffusion
3.1
The main principle
This method indirectly uses the different masses of the isotopes. If a container is filled with gas
and the walls are semi-permeable membranes, then the molecules in the gas will have a small
probability of penetrating the walls. These walls will be referred to as barriers throughout this
chapter. Molecules with a comparatively high speed will encounter the membranes more
frequently than the slower ones and therefore have a larger probability of penetration (see
Section A.4). And since the lighter molecules on average move faster than the heavier ones, the
gas on the other side of the wall will be enriched in the lighter molecules. In principle this process
is very simple, but it is not very efficient because the mass difference between the molecules is so
small. To make the diffusion go faster, a pressure difference is set across the membrane.
The gaseous diffusion technology is relatively simple, but not very effective. The quality that can
make gaseous diffusion interesting to a new participant in the enrichment industry, is that the
technology has been proven to work on an industrial scale. The gaseous diffusion method
employs uranium in the form of uranium hexafluoride gas (see Section A.3).
3.2
The enrichment plant
Each stage in the gaseous diffusion process increases the fraction of U-235 by very little (the
separation factor is about 1.004), but contains a large amount of gas. This makes the process less
flexible. Batch recycling is not a good option for increasing the level of enrichment, but stretching
10
FFI-rapport 2008/02376
Figure 3.1 The gaseous diffusion plant built at Oak Ridge, Tennessee, in the United States during
World War II. Photo courtesy of the United States Department of Energy.
the cascade may be an alternative. If stretching is to be used in a gaseous diffusion plant, then
precautions to prevent the possibility of a criticality accident are necessary to take. This also goes
for batch recycling. More than 3500 stages is needed to produce weapons-grade uranium from
natural feed, and there is an equilibrium time of more than a year [1;3].
3.2.1
The separation stage
The main components of a single gaseous diffusion stage are [4]:
– A large cylindrical vessel called a diffuser or a converter that contains the barrier;
– A compressor used to compress the gas to the pressures needed for it to flow through the
barrier;
– An electric motor to power the compressor;
– A heat exchanger to remove the heat of compression;
– Piping and valves for stage and inter-stage connections and process control.
The key parts in the gaseous diffusion method are the barriers and the compressors. This
technology seems to be relatively simple to develop. It was done by the United States in the
1940s, and reportedly every country that has tried since has been able to do it [5]. When the
gaseous diffusion method was first developed, the corrosive nature of uranium hexafluoride was
one of the major obstacles[1].
There must not be any leakage in the system letting uranium hexafluoride gas out or air in. Metals
such as aluminium and nickel are used for the surfaces that are exposed to the gas, for example
piping and compressors, because they are corrosion resistant. Since there is little enrichment in
just one stage of a gaseous diffusion plant, the cascades are long. Also the equipment tends to be
rather large, so a gaseous diffusions plant takes up a lot of space. When the gaseous diffusion
plant at Oak Ridge was built in the early 1940s (see Figure 3.1), it was the largest industrial
building in the world [6]. Support facilities for a gaseous diffusion plant would include a large
electrical power distribution system, cooling towers, a fluorination facility, a steam plant, a
barrier production plant, and a plant to produce dry air and nitrogen. [4]
FFI-rapport 2008/02376
11
In a typical stage the feed enters at a pressure of one-third or one-half atmosphere. Only half of
the gas is allowed to diffuse through the barriers; this gas will have a lower pressure, and it is sent
through a heat exchanger to be cooled. Then the gas is compressed and mixed with depleted gas
from a higher stage. The cooling is necessary because the compression will substantially increase
the temperature of the gas. This energy-demanding process is the main source of the high power
consumption of the gaseous diffusion process. Modern compressors are similar to those used in
jet aircraft engines. The compressor rotor rotates at high speed and must handle the corrosive
uranium hexafluoride gas at high temperatures in continuous operation for many years. [3]
3.2.2
The diffusion barrier
The barrier must meet some requirements:
– The size of the pores must be very small, but not so small that the diffusion rate gets too low,
and they have to be of a uniform size. The diameter should be much smaller than the average
distance travelled by the molecules between collisions with other molecules (determined by
the pressure of the gas), but still large enough that the molecules are able to penetrate the
barrier at a reasonable rate. About 25 nm would be appropriate if the gas is kept at 80 °C and
a pressure of about one half of an atmosphere [3].
– The barrier must be thin enough for sufficient permeability, but must be thick enough to
withstand the pressure difference across it. The material of the membranes must be bonded
under high temperatures and pressures into thin sheets, and these sheets must be able to
withstand a pressure difference of 0.3 to 0.5 kg/cm2 for a long time (years). This can be
achieved by using many layers of different porosities and strengths [3]. The diffusion barrier
is typically just 5 mm thick [6].
– It must be highly corrosion resistant, so that the holes are not plugged [3]. Typical materials
are nickel and aluminium oxide [4].
The actual methods for producing barriers are kept classified by the countries that use the
technique. Japanese researchers have experimented with nickel, aluminium and teflon barriers.
The United States uses sintered 3 nickel powders while in the French Tricastin plant the barriers
are ceramic. The efficiency of the barriers is about 1 SWU/y per square meter of barrier. [3]
There has been success in using barrier tubes made from anodized 4 aluminium which is usually
considered too fragile for use in an industrial enrichment plant [1].
To increase the efficiency of the method, the surface of contact between the gas and the barrier
should be as large as possible; this can be done by shaping the barriers as thin tubes. The barrier
tubes must be thin to provide a large contact surface with the gas, but not so tight that it hinders
the gas flow. It is assumed that the tubes are about 2 m long and 1 cm in diameter. This means
that 160 000 such tubes are needed in a stage with the capacity of 12 000 SWU/y. [3]
3
Sintering is a process in which a powder is compacted and heated to a point below its melting point; the
particles then attach to each other without melting together.
4
The thickness of the natural oxide layer on the surface is increased; this makes the metal more corrosion
resistant.
12
FFI-rapport 2008/02376
3.2.3
Energy consumption
The gaseous diffusion method consumes a lot of energy, about 2500 kWh/SWU. The energy
consumption was more than ten times this level when the process was first developed, but it
decreased by a factor of ten in about ten years of further development. One method for dealing
with the expenses arising from the high energy consumption, which is inevitable in the gaseous
diffusion method, is to use the plant only in periods of lower electricity prices. This has been done
in France. [1]
3.3
When and where has the technique been used?
The gaseous diffusion method was first used in the United States during World War II, but has
since been used in the United Kingdom, France, China, Russia and Argentina as well. Since
gaseous diffusion is an out-dated and highly power-consuming process, it is not in wide use
anymore. Today only the United States and France use gaseous diffusion on any significant
scale [7].
4
The gas centrifuge method
4.1
The main principle
This method employs uranium in the form of uranium hexafluoride gas (see Section A.3) and
utilizes the mass difference between the isotopes. It is common knowledge that when the only
influence on an object is gravity, it will fall toward the ground. Also if there are several objects,
like grains of sand for instance, in a container they will arrange themselves so that there will be
larger density closest to the ground (the smaller grains will be on the bottom and the lager ones
that cannot be packed as compactly will be on top). This also applies to molecules in a liquid or a
gas. Just like the grains of sand, the molecules of uranium hexafluoride gas made with the heavier
uranium isotope (U-238) will tend to move toward the bottom, and the lighter molecules will tend
to move toward the top because all the molecules take up the same amount of space, and the ones
with the higher mass will therefore also have the higher density. This process is very slow, and
since there is so little U-235 in natural uranium, just filling uranium hexafluoride in a container
and waiting for the molecules to separate into two layers will be very ineffective. Instead it has
proved efficient to construct a fake gravitational field that is much stronger than the earth’s field
by putting the gas in a container and spinning it very fast. The molecules will then be exposed to
the same centrifugal force that we feel when we are driving around a corner, and by spinning the
centrifuge faster, we can make the force stronger, and the separation of the isotopes will happen
quicker. If we use a cylindrical container, we could then in the ideal case extract the enriched gas
from around the centre (the rotation axis) and the depleted gas from near the cylinder wall. But
because the molecules in a gas also perform a random motion (see Section A.4), and all of the gas
will be pressed close to the wall, separating the two isotopes is not so simple, and they cannot be
separated completely.
FFI-rapport 2008/02376
13
Figure 4.1 Sketch of a gas centrifuge. One can see the pattern of the motion of the gas inside the
rotor tube. Here the lower part of the centrifuge is warmer than the upper part so that
the enriched gas will settle in the upper part of the centrifuge. Reprinted with
permission from BBC News.
To increase the separation factor, a relatively slow counter current is set up in the centrifuge. This
is often set up by the same principle that makes it possible to fly a hot air balloon: hot gas rises. If
the temperature is higher in one end of the centrifuge than in the other, the gas in the hot end will
move toward the rotational axis since this is “up” in this “gravitational field” while colder gas
from elsewhere in the centrifuge fills the void, and the gas in the cold end will move toward the
wall since this corresponds to “down.” In this way a convection current flow pattern will emerge
(see Figure 4.1). As can be seen from the figure, the gas closest to the rotational axis will flow
toward the cold end, and the gas closest to the centrifuge wall will flow toward the hot end. Since
the molecules containing the U-235 isotope will be closer to the centre than the molecules
containing the U-238 isotope on average, there will be an accumulation of molecules containing
the light isotope in the cold end, and the gas here will be enriched in U-235. The same flow
pattern can be supported mechanically by use of scoops. Instead of (just) making use of the
temperature difference, the gas is scooped toward the centre in one end and toward the wall in the
other end.
4.2
4.2.1
Technical challenges
Building the centrifuge
From the above discussion, it is obvious that the longer the centrifuge tube and the higher the
rotation velocity, the greater the separation factor. But there are some problems related to
increasing the velocity and the length of the tube that have to be overcome. There will be
vibrations and flexing of the rotor tube as the rotation velocity increases, and this limits the length
14
FFI-rapport 2008/02376
Figure 4.2 Sketch of the different parts of a centrifuge. Copyright © 1992 by the Bulletin of the
Atomic Scientists, Chicago, IL 60637. Reproduced by permission of the Bulletin of
the Atomic Scientists: The Magazine of Global Security, Science, and Survival.
of the tube since it is impossible in practice to balance the tube perfectly. A long thin rotor will
have certain characteristic vibrations, so flexion at the corresponding “resonant” speeds may
cause the rotor or the bearings to break. Therefore passing these “resonant” speeds will be
difficult. Centrifuges which rotate faster than these resonant speeds, and therefore have to go
through them in the acceleration process, are called supercritical. Centrifuges which rotate slower
than the first resonance speed are referred to as subcritical, and are not subject to the same
amount of stresses and strains as the supercritical centrifuges, but are also less efficient. The rotor
tube must be made from a high-strength material, and the vibrations can also be reduced by
reducing the friction force on the rotor. The following will be a discussion of the different
components that are needed to build a centrifuge, and the technical challenges connected to them.
One obvious challenge is the corrosiveness of uranium hexafluoride. The components that come
in direct contact with the gas must be made of or lined with corrosion resistant materials.
As can be seen from Figure 4.2, the centrifuge mainly consists of a rotating tube which contains
the gas. This tube is suspended inside a casing and driven by a motor. The main components of a
centrifuge are:
– A rotor tube
– Bellows
– Top and bottom end caps
These are welded to the rotor tube to contain the uranium hexafluoride. The top cap supports
FFI-rapport 2008/02376
15
–
–
–
–
–
–
part of the upper bearing and the bottom cap carries the rotating part of the bottom
bearing. [8]
An outer casing
A vacuum system
There must be a pumping system to create and maintain vacuum between the rotor and the
casing.
Suspension and bearings
The suspension system keeps the centrifuge in place.
Scoops
The scoops are small tubes which extract enriched and depleted gas, respectively, in either
end of the rotor interior, and they are relatively easy to make. The scoops may be stationary
while the centrifuge rotates around them. They help amplify the convection flow pattern in
the centrifuge and create a pressure difference which makes it possible to extract the product
and tails without using compressors. [1;8]
Motor system
Pipes
Various piping systems are necessary to transport gas of different enrichment into and out of
the centrifuge.
4.2.1.1 Rotor
The maximum speed of the rotor depends on the physical strength of the tube material, and since
most of the stresses on the tube come from its own weight, the tube material must have a high
strength to weight ratio [9]. Suitable rotor materials include alloys of aluminium (alloys
containing zinc, magnesium and copper are the strongest [9]) or titanium, maraging steel 5 or
composites reinforced by certain glass, aramid or carbon fibres. With maraging steel, the
maximum rotor wall speed is approximately 500 m/s, while carbon fibre-resin composites allow
speeds in the range of 700–800 m/s. The aluminium alloys used in early centrifuges permit
maximum speeds of about 350 m/s. Maraging steel is much more difficult to work than
aluminium. Currently, manufacturing carbon fibre-resin composite machines is beyond the
technical capability of states of particular proliferation concern. [10]
The load on the bearings of a spinning centrifuge tube can be reduced by keeping its weight as
low as practically possible, requiring that the walls of the rotor be as thin as possible, typically a
millimetre or so. [9]
4.2.1.2 Bellows
A low-tech solution to the problem of flexural vibrations with long rotors is to break up the length
of the rotor with short, thin flexible joints called bellows, located so as to absorb the energy of the
vibrations at the resonant speeds. While this method is still used in relatively short aluminium and
maraging steel centrifuges, longer, faster carbon fibre-resin composite machines instead use
5
Maraging steel is an alloy of iron containing less carbon than other steels, but a large amount of nickel.
High strength, low weight and corrosion resistance are some of the properties making maraging steel
suitable as a rotor material.
16
FFI-rapport 2008/02376
special drives that can accelerate the rotor rapidly through the resonances before the amplitude of
the vibrations grows sufficiently large to damage the machine. This technology is quite
sophisticated. [10]
The bellows act as a spring, and therefore they make it possible for the rotors to pass the
resonance frequencies without being torn apart. The subcritical centrifuges have short rotor tubes
without bellows [1]. Making the bellows out of maraging steel is one of the difficulties in
manufacturing a centrifuge.
4.2.1.3 Outer casing
The rotor is suspended inside an outer casing, and the space between the rotor and the casing is
evacuated to reduce friction as much as possible. The casing must therefore be air-tight.
Furthermore, the casing also serves the purpose of containing the parts of the centrifuge in case
the centrifuge breaks down, so it must be made of a high-strength material (the rotor moves with
a very high velocity).
4.2.1.4 Suspension and bearings
The suspension system must be capable of damping some amount of vibration because it is
impossible to balance the rotor perfectly to avoid all vibrations [4]. Achieving such precision is
becoming easier as modern, numerically controlled machine tools continue to improve [9].
The top bearings can be made out of ring magnets. This way there does not have to be contact
between the rotor and the suspension on the one hand, and the pipes that carry the uranium
hexafluoride into the rotor on the other hand. The bottom bearing may be a needle with an oil
damper to absorb some of the vibrations. These can be designed such that the rotor acts as a selfstabilizing spinning top. Manufacturing the top and bottom bearings have constituted some of the
greatest challenges to countries trying to build centrifuges. [1]
4.2.1.5 Motor system
AC motors 6 are used, and they provide a centrifuge speed that is proportional to the frequency of
the supplied current. [8] The power supply must accept alternating current input at the 50 Hz or
60 Hz line frequency available from the electric power grid and provide an AC output at a much
higher frequency (typically 600 Hz or more). The high-frequency output from the frequency
changer is fed to the high-speed gas centrifuge drive motors. The centrifuge power supplies must
operate at high efficiency, and provide precise control for the output frequency. [4]
4.2.2
Energy consumption and space requirements
The electrical consumption of a gas centrifuge facility is much less than that of a gaseous
diffusion plant. An energy need around 100–300 kWh/SWU was claimed in 1983 [11], but in
2008 it was claimed to be as low as 50 kWh/SWU [7]. Consequently, a centrifuge plant will not
6
That is, motors that are powered by alternating current (AC).
FFI-rapport 2008/02376
17
have the easily identified electrical and cooling systems typically required by a gaseous diffusion
plant.
A typical centrifuge facility appears to have a capacity of 10–20 SWU/m2, and a facility capable
of producing one bomb per year would then require about 600 m2 of floor space [12].
4.2.3
Feasibility for production of HEU
The throughput of a single centrifuge is small, which leads to rather small separative capacities
for typical proliferator centrifuges. The separation factor of a centrifuge is about 1.5. To be able
to produce only one weapon per year, several thousand centrifuges would be required. [4]
Centrifuges have a small equilibrium time and a small in-process inventory. The inventory is
small because the more gas there is in a centrifuge and the faster it spins, the higher the pressure
near the wall will be. And uranium hexafluoride may then desublime or liquify. Centrifuges are
flexible when it comes to changing the cascade design (parallel to series) [12]. This makes it easer
to change the level of enrichment of the product, and the small equilibrium time and inventory
make batch recycling more feasible. The centrifuge process is economic on a smaller scale, under
2 million SWU per year, which also enables staged development of larger plants. [7]
4.3
When and where has the technique been used?
Today, the gas centrifuge method is taking over for the gaseous diffusion method as the most
important method for uranium enrichment. Plants are being built in all parts of the world, and in
almost all countries that enrich uranium.
The development of centrifuge technology for the separation of isotopes started in the mid 1930s
in Germany. The first known separation of uranium isotopes using a gas centrifuge was
performed in 1941, and the research was conducted by Professor Jesse W Beams and his
colleagues at the University of Virginia. [1;13]
The Soviet Union started developing the gas centrifuge in 1946. The work was conducted in part
by captured German scientists; one of them was Gernot Zippe. In 1956, the Germans were
allowed to leave the Soviet Union. Zippe received permission from the Soviet Minister of Atomic
Energy to work freely on centrifuges in the West in 1958. Zippe contributed to the development
of advanced Western centrifuges and the formation of the uranium enrichment company Urenco.
The first industrial-scale plant in the Soviet Union was built in 1959. [1]
China started research and development of gas centrifuges in 1958, and Chinese officials have
stated that the country may build a gas centrifuge enrichment plant to replace their gaseous
diffusion plants. Russia has agreed to provide China with a gas centrifuge plant. [1]
In Europe, the company Urenco has gas centrifuge facilities in Germany, the Netherlands and the
United Kingdom. Urenco’s present German enrichment plant has been in operation since 1985.
18
FFI-rapport 2008/02376
India also masters the technique of centrifuge separation of isotopes [14], and it is believed that
research on gas centrifuges has taken place in India since the early 1970s or even earlier [1].
The Pakistani scientist Abdul Quadeer Khan was the long-time head of the Pakistani Engineering
Research Laboratories which worked on uranium enrichment after he had worked until 1976 for a
subcontractor to Urenco. He is assumed to have brought with him to Pakistan designs and
assembly instructions for centrifuges and lists of suppliers used by Urenco. The information was
used in Pakistan’s nuclear programme, and has later been sold along with hardware to other
countries. Khan built a whole network trading in enrichment and nuclear bomb technology. Khan
has admitted selling to Libya, Iran and North Korea, but there are suspicions that he has also sold
his services to Iraq, Syria and other countries. [15]
Iraq pursued centrifuge technology in the 1980s. Valuable mechanical design information of an
oil centrifuge were unclassified and available to the Iraqis, and this is what they started working
on. The fact that key parts are moving at great speeds makes centrifuge technology difficult for
inexperienced engineers. In 1988 Iraqi centrifuge experts approached the German firm H&H
Metalform GmbH, for help. H&H was well connected to European centrifuge experts, and the
company became Iraq’s most important contact for assistance in obtaining expertise and
manufacturing equipment form abroad. Iraq received designs for Zippe centrifuges in 1988
through this contact. Iraq was able to import raw materials, motors and frequency converters from
Europe in sufficient quantities for research and development. [1]
Israel may have pursued gas centrifuge technology at least as a research activity, and may also
have expanded on this work [1]. Japan also uses gas centrifuges for uranium production [16].
Urenco received a licence to build a plant in New Mexico, USA, in 2006. This plant was expected
to be on line in 2009 [17]. The United States does not have centrifuge enrichment on an industrial
scale yet, but USEC 7 began construction of a plant in Ohio in 2007. Commercial plant operation
is expected in 2010 [18].
5
Thermal diffusion
5.1
The main principle
The thermal diffusion process is based on the same principle as the counter current flow pattern in
the gas centrifuges, and employs uranium in the form of uranium hexafluoride (see Section A.3).
In this process a thin container of gaseous or liquid uranium hexafluoride is oriented vertically
(see Figure 5.1). One side is then heated and the other side is cooled, and this gives rise to a
thermal convection current with an upward flow along the hot surface and a downward flow
along the cold surface. Under these conditions, the lighter U-235 gas molecules will diffuse
toward the hot surface, and the heavier U-238 gas molecules will diffuse toward the cold surface.
7
United States Enrichment Corporation
FFI-rapport 2008/02376
19
Figure 5.1 Cross section of the film of uranium hexafluoride used in the thermal diffusion
method. Illustration courtesy of the United States Department of Energy.
The result of these motions is that the U-235 molecules concentrate at the top of the film, and the
heavier U-238 molecules concentrate at the bottom. [4]
5.2
When and where has the technique been used?
The thermal diffusion method was one of many methods for enrichment of uranium pursued by
the United States during World War II. It was not found to be a very good method then, and there
is no sign of it having been used since. The process has a low capital cost, but high heating
requirements. It is about 140 times less energy efficient than the gaseous diffusion process [4].
The United States used thermal diffusion to provide slightly enriched feed material for the
electromagnetic separation on the calutrons (see Section 6.1). The film was an annular region
between two cylindrical pipes. The inner pipe was heated by steam, while the outer was cooled by
water. The uranium hexafluoride was kept under pressure to stay in liquid form. The process was
very slow, and only low enrichments were gained even in tubes as high as about 15 m. Still, some
2000 such tubes were constructed to provide feed of 0.89 % U-235 for the calutrons. [19]
6
Electromagnetic processes
The electromagnetic processes for the separation of uranium isotopes all have in common that
they make use of uranium ions 8 . Ions of different mass but equal charge will behave differently in
a uniform magnetic field. The first process, known as EMIS, is relatively simple in theory, and
the main challenge is that it is relatively slow. The two other methods are plasma processes. A
8
Ions have a different numbers of electrons than they have protons; this means that they have a net electric
charge.
20
FFI-rapport 2008/02376
Figure 6.1 Simple drawing of a mass spectrometer. The particles enter in the bottom with the
same kinetic energy and the same electric charge, and the beam will then split in
two because of the different masses of the isotopes. The magnetic field points into
the paper.
plasma is a gas consisting of electrically charged particles. The plasma processes are all fairly
complicated theoretically because the physics of plasmas is rather complicated.
6.1
6.1.1
Mass spectrometer: EMIS (calutron)
The main principle
In a uniform magnetic field, a charged particle with a velocity component orthogonal to the field
will follow a circular or helical path. If it has no velocity component along the field, the
trajectory will be circular in the plane orthogonal to the magnetic field; this is the case in the mass
spectrometer. The radius of the circular path is determined by the strength of the field, the electric
charge, the kinetic energy of the particle and the particle’s mass. A beam of uranium ions with
equal charge and kinetic energy being sent into such a uniform magnetic field will split into two
beams because of the different masses of the isotopes. The beams will form circles, and the one
containing U-238 will have a larger radius than the one containing U-235. This is illustrated in
Figure 6.1. The isotopes can be collected for example after half a rotation in the field. The mass
spectrometer method is often referred to as EMIS (electromagnetic isotope separation), and the
spectrometer is referred to as a calutron, which is what the American machine developed at the
University of California in the early 1940s was called [20].
6.1.2
Equipment and efficiency
In principle, a mass spectrometer consists of a large magnet that can provide a uniform magnetic
field at a sufficient strength, a vacuum pump, an ion source and collectors. The ions of the two
uranium isotopes U-235 and U-238 differ in mass by 1.27 %, and they will have radii that differ
by only 0.64 %. [20]
FFI-rapport 2008/02376
21
The uranium ions can be made from uranium tetrachloride (“green salt”). Solid uranium
tetrachloride is electrically heated to produce uranium tetrachloride vapour. The vapour
molecules are bombarded with electrons, and this knocks loose U+ ions. The ions are then
accelerated over an electric potential to high speeds and follow circular paths in the magnetic
field. Because they are accelerated over the same potential they acquire the same kinetic
energy. [4]
The EMIS process has a very high separation factor (somewhere between 30 and as high as
80 000), but a very low throughput. It also has a very high energy consumption (3800 kWh/SWU
for the early machines, but this value may be lower today). [20]
One limitation is the ion beam intensity because the charges on the ion beam particles will repel
each other. To avoid this problem, the ions can be allowed to collide with a background gas so
that electrons are released and neutralize the repulsion, but if the gas is too dense and the
collisions became too frequent it would disturb the ion beam too much. The typical currents are a
few hundred milliamperes, and at this rate it would take more than 400 years to collect a critical
mass. [20]
Another limitation is the fact that only a small amount of the feed material can be converted into
product in one run. Only about 50 % of the uranium tetrachloride feed is converted to uranium
ions, and only about 50 % of the ions are actually collected in the end. This is a problem if
valuable partially enriched uranium is used as feed, and the insides of the vacuum chamber have
to be cleaned so the material can be recovered. This is a time-consuming process which decreases
the efficiency. [4]
Note that EMIS is an isotope separation method which can be used also to separate the isotopes of
plutonium [20].
6.1.3
When and where has the technique been used?
The method was used, among others, by the United States during World War II to enrich uranium
for their uranium bomb, and the machines were called “calutrons”. Because of the low throughput, over 1000 calutrons were used to help produce uranium for the first American bomb. This
cost roughly a billion dollars over three years [20]. The enrichment was done in two stages; in the
first stage natural or slightly enriched uranium was used to increase the enrichment to 12–20 %;
and in the second stage the product from the first stage was further enriched to weapons grade.
EMIS could be used to produce limited amounts of weapons-grade uranium, and it may be
especially attractive for further enrichment of partially enriched material. The United States, the
United Kingdom, France, the Soviet Union and China all tried this enrichment method. It was
also pursued by the Iraqi uranium enrichment program, and this happened after the process had
been discarded by other countries. The process is both energy demanding and labour intensive,
and it is not economically competitive with other technologies. [4]
22
FFI-rapport 2008/02376
The United States declassified much of its EMIS technology after World War II, and this is well
documented in open literature. Also the basic scientific and technical problems are not too hard to
master, the feed material is relatively easy to produce and handle, and if LEU is used as fuel the
productivity increases substantially. Also, when Iraq decided to use EMIS in the early 1980s, the
computational software and the main equipment were not on any of the international export
control lists, and the prototypes for the main equipment could be designed and manufactured in
the country. Another advantage is that the EMIS technology does not require moving parts at high
speeds but rather static equipment which makes it easier to use for engineers without special
experience. [1]
6.2
6.2.1
Ion cyclotron resonance
The main principle
This process employs the same principle as used in the mass spectrometer, only this time the
particles also have a velocity component along the magnetic field. The motion of the charged
particles becomes a helix with its axis along the magnetic field. The frequency of the rotation is
determined by the strength of the magnetic field and the mass and charge of the particles, and is
called the cyclotron frequency. If the uranium ions have the same charge, then the U-235 ions
will have a larger cyclotron frequency than the U-238 ions. If the ions are subjected to an
oscillating electric field in the plane orthogonal to the magnetic field, which has the same
frequency as the cyclotron frequency 9 of the U-235 ions, then their energy will increase (the ions
are excited) and the radius of the helix for these ions will increase while the radius of the U-238
ions remains about the same. A typical unexcited ion will have a helix radius of the order of a
millimetre, but excited U-235 ion can be made to move in circles several centimetres in diameter
[20]. The ions with the larger radius in their orbit can be collected by plates parallel to the helix
axis while most of the ions which still have a small radius can pass without being caught by the
plates.
6.2.2
How it is done
The ions can be produced from a feed plate of solid uranium. The atoms are vaporized by
bombarding the plate with energetic ions in a process called sputtering. A microwave antenna
located in front of the plate, energizes free electrons which collide with neutral uranium atoms in
the vapour sputtering off the plate. This in turn displaces electrons from the uranium atoms and
produces a plasma of U-235 and U-238 ions. The ions then move in their helical paths toward the
collector inside a cylindrical vacuum chamber. The magnetic field is produced by a super
conducting magnet located around the outside of the chamber. [4]
The strength of the magnetic field is about 2 T which gives the U-235 ions a cyclotron frequency
of 130 kHz. The frequency of the oscillating electric field must be stable to an accuracy of 1 % or
better to distinguish U-235 form U-238. It is difficult to produce a magnetic field which is
uniform to better than 1 % over a large area. Modern magnet technology can achieve this goal
9
The number of rotations in the helix done by the ions per second.
FFI-rapport 2008/02376
23
only with relatively sophisticated and expensive solenoid designs. Because plasmas consist of
charged particles, they shield electric fields. The characteristic shielding length of uranium
plasma is a few micrometres. This means in practice that even though the electric field can be
applied in different ways, it must be applied indirectly. [20]
To separate the two isotopes properly, the ions must be sent through several revolutions. The
influence of the electric field on the U-238 ions will average to zero while the U-235 ions will get
a substantially larger radius. There should be about 80 cyclotron periods before the ions reach the
collector. If the separation tube is 1 m long this means that the axial velocity of the ions cannot be
more than 1 600 m/s which corresponds to a temperature of about 75 000 K. 10 This again means
that the temperature of the plasma has an upper limit for the method to work. When the ions
collide, they will be thrown out of phase with the electric field and loose energy. This puts a limit
on the density of the plasma; the ion density must be below 106 ions per m3. This means that with
a beam diameter of about a metre, using natural uranium as feed and assuming that all the U-235
ions are collected, then about 70 kg of U-235 could be collected per year. [20]
The process needs large amounts of liquid helium 11 to maintain the magnets in their
superconducting state [20].
The process is a batch process, and has to be repeated to produce HEU from natural uranium feed.
The enrichment factor is high, and the energy consumption of the process is quite low, about
330 kWh/SWU. [1;4]
6.2.3
When and where has the technique been used?
The only countries known to have carried out serious experimental programs are the United
States and France. The technique became a part of the Advanced Isotope Separation research and
development program of the American Department of Energy in 1976, but development was
stopped in 1982. The French developed their own version, and this program ended around
1990. [4]
6.3
The plasma centrifuge method
The information in this section is mainly from “Uranium Enrichment and Nuclear Weapon
Proliferation” [20].
In the plasma centrifuge method, a plasma is set into fast rotations. As in a gas centrifuge, the
heavier isotopes will gather closer to the wall of the centrifuge than the lighter isotopes. A main
difference between the gas centrifuge and the plasma centrifuge is that the plasma container does
not have to spin like the rotor in the gas centrifuge. This is a feature of the plasma centrifuge
method which makes it easier to operate at a high rotational velocity. The bearings will not wear
10
The Kelvin temperature scale has the same unit size as the Celsius scale, but it starts at absolute zero.
This means that water freezes at 273 K.
11
Liquid helium has a temperature of 4.2 K (-268.9 °Celsius) at atmospheric pressure.
24
FFI-rapport 2008/02376
out and the walls do not have to withstand the same amount of tension as in the rotating gas
centrifuge. But as the rotational speed increases in the plasma centrifuge, energy consumption and
problems with collisions between ions and neutral particles increase.
The plasma is made to rotate by creating an electric discharge perpendicular to a strong magnetic
field. As in the EMIS and the ion cyclotron resonance processes, the magnetic field makes the
charged particles move in circles. By adding a radial electric field all the particles in the plasma
(independent of whether the charge is negative or positive) will rotate in the same direction and
very high speeds can be attained. In the rotating plasma, the different isotopes will have a velocity
relative to each other. In collisions between ions, the heavier isotopes tend to gain speed and the
lighter to slow down. This increases the velocity difference between the isotopes and enhances
the separation effect.
The energy consumption of the plasma centrifuge method has been said to be about ten times that
for gaseous diffusion, and even though this probably can be reduced, it may still be larger than
that of mechanical centrifuges. One suggestion for reducing the energy consumption is to use a
partially ionized plasma, then the rotation must be transferred to the neutral gas by collisions.
The separation factor of a plasma centrifuge is not too great, but in some designs it may be
possible to enhance it by setting up a counter current flow pattern as in the gas centrifuges.
Several countries have researched plasma centrifuges, also for separation of isotopes of other
elements than uranium. There are a number of variations in the design, but no one has been used
for enrichment of uranium to any degree worth mentioning.
7
Aerodynamic processes
There are two aerodynamic processes that are developed to a degree that makes them possible to
use: The separation nozzle process developed in Germany and the vortex tube separation process
developed in South Africa. They both rely on centrifugal forces to separate the isotopes, but
unlike in the gas centrifuge process there are no spinning parts.
A stream of uranium hexafluoride gas is sent into a curved wall which bends it. During this
bending, centrifugal forces press the heavier gas molecules closer to the wall than the lighter
molecules. Then the stream is separated in one heavier and one lighter fraction.
The centrifugal forces are greater the larger the speed of the gas stream. To make use of this, the
uranium hexafluoride is diluted by a light carrier gas such as hydrogen or helium. The velocity of
sound is higher in a lighter gas than in a heavy gas, and this means that the gas can travel faster
without energy being lost in sonic shock effects. Also the light carrier gas makes it easier for the
isotopes of uranium to separate.
FFI-rapport 2008/02376
25
Figure 7.1 Schematic view of a separation nozzle. The smaller the nozzle is, the less turbulence
there will be in the gas flow. The sketch is reprinted here with permission from
Informationskreis KernEnergie, Berlin (www.kernenergie.de).
Both of the aerodynamic processes described below have a high energy consumption because of
the high proportion of carrier gas relative to uranium hexafluoride. The processes are not
considered economically competitive because of this high energy consumption.
A complicating feature of the aerodynamic processes is the handling of hydrogen. If the
temperature is too high, hydrogen will react with uranium hexafluoride, so the temperature must
be kept under control. [1;4]
7.1
The separation nozzle process
The separation nozzle process, also known as the Becker nozzle process after its inventor, was
developed in Germany and has been used both there and in Brazil, but both countries have since
abandoned this activity [1]. Gaseous uranium hexafluoride diluted by hydrogen or helium is
compressed and directed along a curved wall at high speed, and the heavier molecules gather
closer to the wall. The centrifugal forces on the gas in the nozzle can exceed those in a centrifuge
by more than a thousand times [21]. Then the stream is split by a knife edge, and the heavy and
the light fraction is withdrawn separately [4] (see Figure 7.1).
The smaller the physical dimension of the separation nozzle, the more economic it is, and the
radius of curvature may be as small as 10 µm. Production of the nozzles is done by photo etching
of metal foils and then stacking them. A stage typically consists of a vertical cylindrical vessel
containing the separation elements, a cross piece for gas distribution, a gas cooler to remove the
heat of compression and a centrifugal compressor driven by an electric motor. [4]
The energy costs for a separation nozzle plant is about 3600 kWh/SWU, and originate mostly in
the re-compression of the depleted and enriched gas streams. The separation factor of this method
is bout 1.015 which is lower than for gas centrifuges but higher than for gaseous diffusion [1;22].
26
FFI-rapport 2008/02376
7.2
The vortex tube process
The process developed in South Africa can be characterized as an advanced vortex tube or a
stationary-wall centrifuge. In this process, uranium hexafluoride gas is diluted by hydrogen gas
and then compressed. The gas mixture enters tangentially through nozzles or holes at speeds close
to the speed of sound into a tube shaped like a cone [4] The gas spirals down the tube and at the
end of the tube the heavier gas exits through holes in the side while the lighter gas exits straight
out the end [1]. The vortex tube has a relatively high separation factor of between 1.025 and
1.035 [23;24].
The South Africans developed a technique called Helikon to cascade the elements, and avoid
difficulties in piping. A typical Helikon module consists of a large cylindrical steel vessel that
houses a separation element assembly, two axial-flow compressors (one mounted on each end)
and two water-cooled heat exchangers. [4] A module with the capacity of 10 000 SWU per year
would be about 3.6 m in diameter and 10 m long.
South Africa used the vortex tube method to produce both reactor fuel and HEU for their
weapons program [25]. The Y-plant at Valindaba produced hundreds of kilograms of HEU [4].
This plant was the pilot enrichment facility designed to produce weapons-grade uranium, and it
started operation in 1977. The first enriched uranium was withdrawn in 1978. In 1988 production
started at another plant located next to the Y-Plant. This plant was semi-commercial and
employed the Helikon technique, and it produced 3.25 % enriched uranium for a power reactor.
The plant could be used to produce HEU by batch recycling or stretching, but was not intended
for this.
The energy consumption of the vortex tube process is about 4000 kWh/SWU, which is high
compared to other techniques. Because of the high costs of this process, South Africa also
investigated both gas centrifuge and MLIS processes (see below). [1]
8
Laser isotope separation
Laser isotope separation makes use of the quantum mechanical properties of atoms and molecules
described in Section 10A.5. Lasers are light sources that can provide an intense beam of light
with a very precise wavelength or energy of the photons. If one can produce a laser capable of
emitting light at the exact frequency that is needed to separate the two isotopes of uranium, it will
be a very efficient way of enriching.
Although the laser isotope separation (LIS) techniques are relatively simple to understand, they
are considered technically difficult, and it has been assumed that only technically advanced
countries would be able to master them. Much relies on laser and optical devices. The laser
industry is in rapid development on its own, and it is therefore not unlikely that the technical
obstacles will decrease as time passes. LIS can of course be used on other chemical elements than
uranium, so the technology is interesting also for other purposes than uranium enrichment. It is
FFI-rapport 2008/02376
27
Figure 8.1 Ionization of the uranium atom is done in three steps of about equal size. Each arrow
in the diagram represents one laser that has to be present for the AVLIS process to
work. The energy levels shown are quoted from [26].
therefore possible to “practise” the use of LIS technology and LIS equipment without using
uranium just by tuning the lasers to a different frequency.
Unlike many other isotope separation methods, LIS can be used to separate the isotopes of
plutonium [26].
8.1
8.1.1
Atomic vapour laser isotope separation (AVLIS)
The main principle
The atomic vapour laser isotope separation (AVLIS) process consists of a laser system and a
separation system. Uranium in its pure metallic form is melted and heated in a crucible by an
electron beam to the point where enough of the atoms evaporate. The evaporated atoms move
radially away from the ingot. Warm liquid uranium is very corrosive and must not come into
contact with the support structure. It is also important that the heat does not disappear, and that
the atoms can move relatively freely; therefore an extensive mechanism is needed to create
vacuum and contain the heat. The liquid uranium is in one end of an evacuated chamber, and laser
beams are cutting through the other end. As the uranium gas moves away from its source, the
atoms collide and most of their internal energy is converted to kinetic energy. When the atoms
reach the radiation zone (the laser beams), about 40 % of them are in their lowest energy state
(the ground state). [26]
When the atoms reach the radiation zone, the U-235 atoms which are in the ground state will be
excited two times and then ionized (see Figure 8.1). The ionization energy for the uranium atom
is about 6.2 eV. 12 In theory, just one laser with the exact energy could be used to ionize the atom,
12
The electron volt (eV) is a small physical unit for energy, appropriate on atomic levels. It is the same
amount of energy as 1.602 x 10-19 joules.
28
FFI-rapport 2008/02376
but in practice it is necessary to do the ionization in at least two steps. Three steps of about equal
size have been used to take advantage of the readily accessible and well-established technology of
rhodamine dye lasers (see Appendix B for more information on lasers). The ions are then
collected on charged plates oriented parallel to the radial direction, while most of the neutral
atoms continue outwards and are collected on plates orthogonal to their direction of motion. It is
impossible to avoid collecting some neutral atoms on the collector plates for the ions because of
the random motion of the molecules in a gas (see Section A.4). This and other problems decrease
the efficiency such that the maximum separation factor will be about 15. In practice, a separation
factor of between five and ten is more likely, and this is far greater than the separation factor of
processes that are already in use, such as gaseous diffusion and gas centrifuges where the
separation factors are just above 1. [26]
An AVLIS module consists of the crucible holding the liquid uranium and the collector plates,
and it is encapsulated in a vacuum system. All laser sources and mirrors are kept outside the
volume containing the uranium steam. Each module can be typically 1 m high and 3 m long. The
laser light is reflected many times by mirrors and may have a path length of more than
100 m. [26]
Since only about 40 % of the U-235 atoms are in their ground state, only these 40 % can be
excited, and this lowers the efficiency of the process. But an additional about 30 % are in a low
excited state and can be included by adding another laser with a slightly lower energy which
excites these atoms into a state with the same energy as the atoms in the ground state are excited
to (cf. Figure 8.1). In this way, 70 % of the U-235 atoms may be ionized instead of only 40 %.
The four lasers that are needed all have slightly different colours corresponding to the slightly
different energies required, but all are in the red-to-orange region of the spectrum. [26]
8.1.2
Advantages and problems with the AVLIS process
One of the advantages of the AVLIS process is the high separation factor. This suggests that
producing highly enriched uranium would be easier with this process than with many other
processes. With a separation factor of ten, only three times of batch recycling will be needed to
reach a product of 97 % U-235. But there are some obstacles. When the fraction of U-235 in the
feed is increased, a higher laser intensity is needed to maintain the same separation factor because
there are more atoms present which need to be excited. When the feed contains more than 50 %
U-235, it will be better to tune the lasers to ionize the U-238 atoms. Then the intensity of the
lasers will not have to increase as the enrichment level becomes larger, but can instead decrease.
If the steam is too dense, it will be more difficult to collect alt the atoms, and some will be lost.
On the other hand, if the density of the steam is to low, the atoms will not have the opportunity to
collide often enough to reach the ground state before they enter the radiation zone. [26]
Also only about 50 % of the atoms reach the radiation zone with the laser beams after being
evaporated, the remaining atoms are collected on the walls of the chamber and must be recycled.
This makes the process less efficient. [26]
FFI-rapport 2008/02376
29
The AVLIS enrichment process will generate a small amount of waste and consume only about as
much energy as the centrifuge process, 100–200 kWh/SWU [4;23]. The process does not require
as much space as a centrifuge plant, however [27].
8.2
8.2.1
Molecular laser isotope separation (MLIS)
The main principle
The energy differences between the vibrational states of a molecule are smaller than the
differences between the electronic states of an atom, so for the dissociation of molecules infrared
light must be used instead of visible light. Photo dissociation of a molecule (see Section A.5) can
be done in one, two or more steps. When two steps are used in the molecular laser isotope
separation (MLIS) process for enrichment of uranium, the first photon is in the infrared section of
the electromagnetic spectrum. This is the one that selects the desired molecule by exciting the
vibrational states of the molecules containing U-235 and not U-238. Then a photon of higher
energy, in the ultraviolet end of the spectrum, breaks the bond completely. Molecules which are
already excited have a larger probability of absorbing the photon and be dissociated than those in
the ground state, and in this way the selectivity of the first step is conserved. But it is not
impossible also for the unexcited molecules to be dissociated by the UV photons, and this makes
the process less effective. In this two-step process the excitation and the dissociation must occur
on a time scale that is short compared to the lifetime 13 of the intermediate vibrational state.
Alternatively, several infrared photons of the same frequency may be used to dissociate the
molecules and still be isotopically selective. [28]
The MLIS process uses uranium hexafluoride gas because of its chemical qualities 14 , and because
its handling is well known from its use in the gaseous diffusion method and the gas centrifuge
method. Other molecules could also be used. The idea is to break the bond between the uranium
atom and one of the fluorine atoms in the molecules containing U-235 so that uranium
pentafluoride and a free fluorine atom are formed. Uranium pentafluoride enriched in U-235 will
then precipitate 15 as a white powder and can subsequently be collected [4].
As in the AVLIS process, it is the ground state that has to be occupied initially if the laser is
going to be able to excite the molecules. The molecules in the uranium hexafluoride gas collide so
often that almost none of them occupy the ground state. The molecules can be forced into the
ground state by cooling, but before room temperature is reached uranium hexafluoride will
precipitate. One way of getting around this is by super-cooling the gas. Super-cooling means
cooling below the sublimation point without crystallization taking place. This is a very unstable
state, and if just a small crystal is present or the gas is disturbed, then crystallization will occur.
Gaseous uranium hexafluoride mixed with a carrier gas and a scavenger gas is expanded through
a nozzle at large speed, this cools the gas down to low temperatures without the uranium
13
The lifetime is the average time it takes from a molecule is excited until it deexcites by getting rid of its
excitation energy.
14
One being the possibility to convert it into uranium pentafluoride by photo dissociation.
15
Form solid particles.
30
FFI-rapport 2008/02376
hexafluoride precipitating. Suitable carriers are hydrogen or a noble gas. A scavenger gas (such as
methane) is used to capture the fluorine atoms that are released as a result of the dissociation of
the uranium hexafluoride molecules containing U-235. [4]
By this process, 95 % of the molecules will descend to the ground state, and only one laser is
needed to excite most of the molecules (unlike in the AVLIS process where two lasers were
needed and only 60 % of the atoms were excited). [26]
The energy consumption of the MLIS process is estimated at about 235 kWh/SWU which is
slightly higher than for the AVLIS process. The use of uranium hexafluoride and low power
consumption are some of the advantages of the MLIS process [1;4].
8.2.2
Potential for HEU production
It was a concern when developing the process that the uranium pentafluoride and the uranium
hexafluoride molecules might exchange fluorine atoms before being separated, but this turned out
not to be a big problem [28]. Also as the enrichment in the feed increases, this problem will
decrease because there will be fewer U-238 atoms present to change places with the U-235 atoms.
This makes the MLIS process more suitable for cascading than the AVLIS process. Also
removing the uranium pentafluoride and converting it back to uranium hexafluoride does not pose
great difficulties. And the amount of U-235 can always be made to match the lasers, independent
of the enrichment level, because the uranium hexafluoride gas is diluted by a carrier gas. All of
this implies that using the MLIS process for HEU production may be possible in relatively small
plants.
8.2.3
Separation of isotopes by laser excitation (SILEX)
The SILEX process was developed in Australia by the company Silex Systems Limited (Silex). It
is an MLIS process using uranium hexafluoride as feed [29]. Many details of the process are
classified, and it is difficult to distinguish SILEX from the general MLIS technique described
above. The technique uses a pulsed CO2 laser of wavelength 10.8 µm which is converted to about
16 µm by use of a parahydrogen Raman converter [30]. The 50 Hz pulse rate of the lasers is not
enough to produce significant enrichment; it must be increased at least ten times [30]. Both the
product and the tails are removed mechanically, and the separator system is cascaded [29].
SILEX has advantages to other processes for uranium enrichment such as low power
consumption, simple and practical separation modules and versatility. [31] Because of this, the
SILEX method is suggested as the third generation of uranium enrichment, where gaseous
diffusion and gas centrifuges constitute the first and second generations, respectively. The
separation factor is classified, but announced by the company Silex to be somewhere between 2
and 20 whereas it is 1.3 for gas centrifuges and 1.004 for gaseous diffusion [31].
FFI-rapport 2008/02376
31
8.3
Laser assisted processes
The information in this section is from “Uranium Enrichment and Nuclear Weapon
Proliferation” [26].
The laser’s ability to excite only one kind of molecule can be utilized in several different ways in
addition to those mentioned above. One example is that they may raise the temperature of one
component of a gas mixture so rapidly that the rest of the mixture is not heated notably during the
process. This could be done to uranium hexafluoride molecules by use of an infra red laser similar
to the one used in the MLIS process. When the molecules are excited into a higher vibrational
state, and the molecules collide with others, much of the vibrational energy can be converted into
kinetic energy, which makes the molecules move and diffuse faster. This is a quality that can be
used to enhance the effect of some of the methods which use uranium hexafluoride.
One example is the thermal diffusion process. If only uranium-235 hexafluoride molecules are
heated, then fewer stages are needed to achieve the same separation of U-235 and U-238 than
without the use of lasers. Another method is to combine it with an aerodynamic process.
Another proposed method involves letting a stream of a gas mixture of uranium hexafluoride and
argon be irradiated by an infrared laser. At the end of the tube, there are very cold plates (below
-150 °C) on which the gas molecules will condense if they collide with it. Because the molecules
containing U-235 will have a higher velocity, they are more likely to collide with the plates.
When the tube is evacuated, the plates can be heated to free the uranium hexafluoride in gaseous
form again and the gas can be sent on to the next stage in the process.
8.4
When and where have the LIS techniques been used?
More than 20 countries are known to have done research on laser isotope separation techniques. 16
France, the United Kingdom and the United States have developed LIS programs that could move
beyond the lab into the pre-industrial phase and ultimately into commercial production. In August
2002, it was announced by the dissident group National Council of Resistance of Iran that Iran
had started an LIS program and developed a laser enrichment facility. This enrichment program
was not detected by the International Atomic Energy Agency (IAEA), but had supposedly only
enriched milligrams of uranium. In September 2004, IAEA discovered South Korean
experiments; in 2000, scientists at the Laboratory for Quantum Optics at the Korea Atomic
Energy Research Institute had separated about 0.2 g of U-235 enriched to levels between 10 %
and 77 %. It seems that LIS methods are hard to upgrade to commercial scale, but scientists
would not need a commercial-scale plant to enrich enough uranium for a single nuclear weapon if
they had one or two years in which to do it. [32]
16
This includes Argentina, Australia, Brazil, China, France, Germany, India, Iran, Iraq, Israel, Italy, Japan,
the Netherlands, Pakistan, Romania, Russia, South Africa, South Korea, Spain, Sweden, Switzerland, the
United Kingdom, the United States and Yugoslavia.
32
FFI-rapport 2008/02376
Figure 8.2 Laser corridor during AVLIS operation. Over 2000 optical elements convey these
green and yellow beams. Each wall in the dye laser corridor has two dye laser
chains, which amplify the light for later use in the separator tank. Special filters
clean the air in the laser enclosures, making it 1000 times cleaner than that in a
room with standard air conditioning. Photo courtesy of the United States
Department of Energy.
Figure 8.3 Photograph showing some of the approximately 4500 optical elements used in the
AVLIS process. These high-quality optical elements reflect light approximately
100 times more efficiently than a bathroom mirror. Optical elements convey light
through the laser system, and some also regulate the size and shape of the laser
beams. Photo courtesy of the United States Department of Energy.
The challenges posed by the AVLIS technique are mainly technical. The complex laser systems
(cf. Figure 8.2 and Figure 8.3) and the fact that uranium vapour is very corrosive, lead to
technical difficulties that make the process hard to apply on a commercial scale. In June 1999, the
United States was on the verge of commercialization when it was decided to cancel its AVLIS
program. About two billion dollars had then been spent on the program. Faced with stiff
FFI-rapport 2008/02376
33
competition from gas centrifuges, large-scale LIS processes have not been developed by any of
the countries who have sought this path for uranium enrichment. [32] The only potential largescale LIS project seems to be the one conducted by the Australian company Silex (cf.
Section 8.2.3).
Silex Systems Limited (Silex) was established in 1988. The principle was demonstrated to work
in 1994, and in 1996 Silex reached an agreement for the development and licensing of SILEX
technology with the United States Enrichment Corporation (USEC). In 2000, the process was
demonstrated at prototype level, and in 2001 the technology was classified by the United States
and Australian governments. The classification of the process has been interpreted by some to
mean that the process can be used to produce highly enriched uranium for weapons use, but in
2007, it was claimed that the SILEX process was not mature enough to enrich significant amounts
of uranium to the levels required in nuclear weapons [30]. USEC withdrew from the project in
2003, but in May 2006, General Electric Company signed an exclusive commercialisation and
license agreement for the SILEX uranium enrichment technology. [33]
9
Chemical separation methods
Two chemical processes have been developed to a point where they could be considered possible
to use for the enrichment of uranium. The CHEMEX process was developed in France, and an ion
exchange method has been developed in Japan. The chemical approaches to the separation of
uranium isotopes are attractive because of their low energy consumption; there is no need for
powerful compressors or pumps [34]. Furthermore, these processes are based on standard
chemical engineering technology [4], which makes them favourable.
Both processes are based on the fact that U-235 is more likely than U-238 to be in a higher
oxidation state (see Section A.6) [4]. This means that if natural uranium is present in different
chemical states or compounds, then there will be an accumulation of U-235 in the higher
oxidation states. If it is possible to separate the states (for example if they are different types of
molecules that do not mix very well) then the compound which is enriched in U-235 can be
extracted. For the separation effect to be the greatest, the two compounds used should have
uranium in as different oxidation states as possible, and the temperature should be as low as
possible [34]. Unfortunately these are criteria which also make the process go slower and lead to
the need of good catalysts 17 .
A separation stage consists of mixing of the two compounds of uranium for a long enough time
that chemical equilibrium may be established. This can be done by letting the two substances
flow in opposite directions in a column. The main difficulty of chemical enrichment methods is
the reflux problem. When the substances emerge at the ends of the column, some can be extracted
as product and tail, but much has to be sent back into the column to maintain the counter current
flow pattern. Before they can be sent back into the column, the enriched substance has to be
17
A catalyst is a substance which facilitates a chemical process without being consumed itself.
34
FFI-rapport 2008/02376
transformed into the form of the depleted substance and vice versa. This is a non-reversible
process and the main source of power consumption in the chemical enrichment processes. Even a
small material loss in the reflux processes can reduce the efficiency of the plant significantly. [34]
9.1
The French process (CHEMEX)
In the French process, two different compounds including uranium are dissolved in different
liquids. These liquids are immiscible 18 in order to make possible the later separation of the
compounds; one is in an aqueous phase, and one is in an organic phase. [22]
The two liquids flow through each other in a column. The liquids are agitated so that the organic
one forms small droplets to increase the reaction surface; this is done either by the use of an
external mechanical device or a series of reciprocating discs. Limiting factors are the rate of the
chemical reaction and the rate of diffusion in and out of the oil droplets for the two chemical
species. The chemical reaction rate can be sped up by changing the uranium compounds or by the
use of catalysts and raising the temperature. Some compounds from the group known as “crown
ethers” have good qualities for uranium enrichment because they both extract the uranium from
one liquid phase to another and enhance the separation effect. The diffusion rate of the uranium
compounds in and out of the oil droplet can be made higher by decreasing the size of the droplet.
But there seams to be a practical limit at a diameter of a few millimetres, and this makes the
process slower than the Japanese process discussed below (Section 9.2). [34]
The French enrichment facility is proposed to be constructed of two identical columns connected
in a way that reduces the equilibrium time. The standard dimensions of the exchange column in
the CHEMEX process is 1.6 m in diameter and 35 m high [22], and the capacity is
3000–5000 SWU per year. The concentration of uranium is greater in this process than in the
Japanese process. The inventory is about one order of magnitude larger than that for a gaseous
diffusion plant of about the same capacity. There is also a need for large volumes of chemicals.
Some of them, such as the crown ethers, can be expensive. [34]
At the ends of the column, the two uranium compounds have to be transformed into each other, in
order to have reflux. The chemical converters must be very efficient; they are the main sources of
energy consumption in this process. [22]
The uranium compounds, the catalysts and the reflux reactions are kept secret. The components
needed in the process are made from plastic materials because of corrosion problems. [34]
The estimated energy consumption is about 600 kWh/SWU [23].
9.2
The Japanese process (ACEP)
The Japanese ion exchange process known as the Asahi Chemical Exchange Process (ACEP) has
been under development since 1972. It is based on a solid uranium adsorbent which has been
18
Immiscible liquids do not mix. An example is oil and water.
FFI-rapport 2008/02376
35
developed by Asahi Chemical; this selectively adsorbs 19 hexavalent uranium. The breakthrough
of this process was the discovery of a uranium adsorber of high reaction speed. [22] The adsorber
is a resin 20 in which the individual particles have the forms of spheres with a porous surface to
increase the contact surface. It has the function of holding one of the two uranium compounds on
its surface [34].
First the resin is charged with an oxidizing agent. A solution containing a uranium compound in
which uranium is in the 4+ oxidation state is introduced. Then the uranium is oxidized to the 6+
state and replaces the oxidizing agent on the resin. This process will leave a sharp boundary
between the part of the column where the resins are occupied by the oxidizing agent and uranium,
this boundary will move slowly down the column when more uranium solution is added at the top
of the column. After some time a new solution is added at the top of the column, this one
containing a reducing agent. This will act on the uranium absorbed on the resin and reduce it back
to the 4+ oxidation state and then replace it on the resin. The 4+ uranium goes into the solution
and continues down the column with it. The 4+ uranium will then come into contact with the 6+
uranium on the resins further down the column, and the isotopic change will take place. There
will eventually be a higher concentration of U-235 in the compound adsorbed on the resin than in
the solution. The portion of the column where the isotope exchange takes place is well defined
because also the reducing agent forms a sharp boundary with the uranium compound. [34]
The process is carried out by the use of four columns; the actual ion-exchange process takes place
in the first three columns which are connected in series, and the recharging of the agents is done
in the fourth. Each column should be a few meters high. Very little energy is needed in the ionexchange columns unless the temperature has to be raised. The main energy consumption comes
from recharging the oxidizing and reducing agents and the pumping of the recycled solution. [34]
The energy consumption of the process is estimated to be about 150 kWh/SWU [23]. Because of
good exchange kinetics, the ACEP process requires smaller volumes than the CHEMEX process
[22].
To find suitable resins, oxidizing and reducing agents, processes for recharging these as well as a
suitable catalyst for the exchange process has taken many years. [34]
9.3
A third process
There is also a third process that has been studied by several countries. It appears to be attractive,
but there are problems related to developing it past the laboratory stage. The problems seem to be
mainly large energy costs and losses of material in the recycling of the compounds. The
development of this process seems to be far behind the two processes discussed above. It makes
use of NOUF6 dissolved in dry hydrogen fluoride and UF6 dissolved in freon or some other
saturated fluorocarbon in the liquid phases. It is the execution without large energy costs and
19
When something is adsorbed it forms a layer on the surface of the adsorber.
A resin is a substance that acts as an ion exchanger. It holds ions in its structure and is therefore able to
hold ions with the opposite charge to its surface. When new ions are adsorbed, old ones are released.
20
36
FFI-rapport 2008/02376
losses of material of the reaction that recycles NOUF6 which seems to be the biggest
challenge. [34]
9.4
When and where has the technique been used?
Chemical exchange processes are under study in several countries, but there are still no full-scale
chemical enrichment plants in the world [4]. The only countries that have attempted this
technique to any extent worth mentioning seem to be France, Japan and Iraq. The French process
is offered on sale to countries that do not pursue other enrichment paths [5]. It is believed that
North Korea may have done research on chemical enrichment using knowledge obtained in Japan
[35;36].
Iraq decided to concentrate on chemical enrichment methods for the production of LEU to be
used as fuel for their EMIS program when their gaseous diffusion program was not progressing
well in the late 1980s [1].
10 Brief summary and comments
Even though a number of methods for the enrichment of uranium have been investigated and
developed over the years, only very few have been applied on a large scale, and presently the gas
centrifuge method appears to be the only commercially viable method. It is possible that other
techniques, such as for example laser isotope separation, may be successfully used also on an
industrial scale in the future.
One should be aware that some of the techniques that cannot be used for large-scale enrichment,
for example the mass spectrometer method, may still be of use to small-scale bomb makers. It is
therefore important to maintain export control regimes capable of preventing equipment for all
kinds of uranium enrichment from falling into the wrong hands.
References
[1] D. Albright, F. Berkhout, and W. Walker, Plutonium and highly enriched uranium 1996
world inventories, capabilities and policies., Oxford University Press, 1997.
[2] A. S. Krass, P. Boskma, B. Elzen, and W. A. Smit, "General principles of uranium
enrichment," in Uranium Enrichment and Nuclear Weapon Proliferation Taylor and
Francis Ltd, 1983.
[3] A. S. Krass, P. Boskma, B. Elzen, and W. A. Smit, "Enrichment processes, II Gaseous
diffusion," in Uranium Enrichment and Nuclear Weapon Proliferation Taylor and
Francis Ltd, 1983.
[4] Department of Defense, "The Militarily Critical Technologies List Part II: Weapons of
mass destruction technologies, Section 5.2 - Uranium enrichment processes," pp. II-513-II-5-19, Department of Defense, Office of the Under Secretary of Defense for
Acquisition and Technology,1998.
FFI-rapport 2008/02376
37
[5] S. Tyson, "Uranium enrichment technologies: Proliferation implications," Nuclear Threat
Initiative, www.nti.org/db/archives/nuc/eos/tyson.htm, last update 1999, accessed 2008.
[6] A. Makhijani, L. Chalmers, and B. Smith, "Uranium enrichment Just Plain Facts to Fuel
an Informed Debate on Nuclear Proliferation and Nuclear Power," Oct. 2004.
[7] World Nuclear Association, "Uranium Enrichment," World Nuclear Association,
http://www.world-nuclear.org/info/inf28.html, last update 2008, accessed 2008.
[8] D. Albright and M. Hibbs, "Iraq's shop-till-you-drop nuclear program," Bulletin of the
Atomic Scientists, vol. 48, no. 3 1992.
[9] I. Oelrich, "All Spun Up: Tutorial On Gas Centrifuges' Role in Proliferation," Federation
of American Scientists, http://www.fas.org/programs/ssp/nukes/fuelcycle/allspunup.html,
last update 2008.
[10] V. Gilinsky, M. Miller, and H. Hubbard, "A fresh examination of the proliferation
dangers of light water reactors, Appendix 1," Nonproliferation Policy Education Center,
http://www.npec-web.org/Essays/20041022-GilinskyEtAl-LWR.pdf, last update 2004.
[11] A. S. Krass, P. Boskma, B. Elzen, and W. A. Smit, "Enrichment processes, III The gas
centrifuge," in Uranium Enrichment and Nuclear Weapon Proliferation Taylor and
Francis Ltd, 1983.
[12] Global Security, "Gas Centrifuge Uranium Enrichment," Global Security,
http://www.globalsecurity.org/wmd/intro/u-centrifuge.htm, last update 27 Apr. 2005,
accessed 2008.
[13] R. Green, "Back to the future," Nuclear engineering international, no. September 2003,
pp. 36-39, 2003.
[14] D. Albright and S. Basu, "India's Gas Centrifuge Enrichment Program: Growing Capacity
for Military Purposes," Institute for Science and International Security,Jan. 2007.
[15] H. Breivik and H. K. Toft, "The A. Q. Kahn network," Norwegian Defence Research
Establishment (FFI), 2007/00535, Feb. 2008.
[16] Japan nuclear fuel limited, "Uranium Enrichment," Japan nuclear fuel limited,
http://www.jnfl.co.jp/english/uranium.html, last update 2006, accessed 5 Aug. 2008.
[17] Urenco, "National Enrichment Facility, About LES," Urenco,
www.urenco.com/fullArticle.aspx?m=1707, last update 2006, accessed 2008.
[18] USEC, "The American Centrifuge," United States Enrichment Corporation,
http://usec.com/americancentrifuge.htm, last update 2008, accessed 7 Aug. 2008.
[19] A. S. Krass, P. Boskma, B. Elzen, and W. A. Smit, "Enrichment and proliferation," in
Uranium Enrichment and Nuclear Weapon Proliferation Taylor and Francis Ltd, 1983, p.
14.
[20] A. S. Krass, P. Boskma, B. Elzen, and W. A. Smit, "Enrichment processes, VII
Electromagnetic and plasma processes," in Uranium Enrichment and Nuclear Weapon
Proliferation Taylor and Francis Ltd, 1983.
38
FFI-rapport 2008/02376
[21] A. S. Krass, P. Boskma, B. Elzen, and W. A. Smit, "Enrichment processes, IV
Aerodynamic separation methods," in Uranium Enrichment and Nuclear Weapon
Proliferation Taylor and Francis Ltd, 1983.
[22] S. Villani, "Progress in uranium enrichment," Naturwissenschaften, vol. 71, no. 3, pp.
115-123, 1984.
[23] D. Albright, F. Berkhout, and W. Walker, "Weapon-grade plutonium and highly enriched
uranium production," in Plutonium and highly enriched uranium 1996 world inventories,
capabilities and policies Oxford University Press, 1997, p. 468.
[24] M. A. Fordt, "Uranium in South Africa," Journal of The South African Institute of Mining
and Metallurgy, vol. 93, no. 2, p. 54, 1993.
[25] FAS, "Uranium Production," Federation of American Scientists,
www.fas.org/nuke/intro/nuke/uranium.htm, last update 24 June 2000, accessed 1 Aug.
2008.
[26] A. S. Krass, P. Boskma, B. Elzen, and W. A. Smit, "Enrichment processes, VI Laser
Isotope Separation," in Uranium Enrichment and Nuclear Weapon Proliferation Taylor
and Francis Ltd, 1983.
[27] A. S. Krass, P. Boskma, B. Elzen, and W. A. Smit, "Enrichment and proliferation, III
Comparison of techniques," in Uranium Enrichment and Nuclear Weapon Proliferation
Taylor and Francis Ltd, 1983.
[28] R. J. Jensen, O. P. Judd, and J. A. Sullivan, "Separating Isotopes With Lasers," Los
Alamos Science, vol. 3, no. Number 1 Winter/Spring 1982.
[29] H. Jones-Bey, "LASER ISOTOPE SEPARATION: Fuel-enrichment method garners GE
contract," Laser Forcus World, http://www.laserfocusworld.com/articles/266374, last
update 1 Aug. 2006, accessed 4 Aug. 2008.
[30] J. Hecht, "LASER ISOTOPE SEPARATION: Silex uranium-enrichment scheme may not
be mature enough for practical use," Laser Forcus World,
http://www.laserfocusworld.com/display_article/308728/12/none/none/News/LASERISOTOPE-SEPARATION:-Silex-uranium-enrichment-schene-may-not-be-matureenough-for-practical-us, last update 1 Aug. 2007, accessed 4 Aug. 2008.
[31] Silex, "Silex - Separation of Isotopes by Laser EXcitation," Silex,
http://www.silex.com.au/s03_about_silex/s30_1_content.html, last update 2008, accessed
28 July 2008.
[32] J. Boureston and C. D. Ferguson, "Laser enrichment: Separation anxiety," Bulletin of the
Atomic Scientists, vol. 61, no. 2, pp. 14-18, 2005.
[33] Silex, "Company History," Silex,
http://www.silex.com.au/s02_company_history/content.html, last update 2008, accessed
28 July 2008.
[34] A. S. Krass, P. Boskma, B. Elzen, and W. A. Smit, "Enrichment processes, V Chemicalexhange methods," in Uranium Enrichment and Nuclear Weapon Proliferation Taylor
and Francis Ltd, 1983.
FFI-rapport 2008/02376
39
[35] D. Albright, F. Berkhout, and W. Walker, "North Korea," in Plutonium and highly
enriched uranium 1996 world inventories, capabilities and policies Oxford University
Press, 1997.
[36] D. Albright, F. Berkhout, and W. Walker, "A special case: Iraq," in Plutonium and highly
enriched uranium 1996 world inventories, capabilities and policies Oxford University
Press, 1997.
[37] R. Vandenbosch and J. R. Huizenga, Nuclear Fission., Academic Press, 1973.
[38]
CRC Handbook of Chemistry and Physics, 66 ed., CRC Press, 1986.
[39] DoE, Department of Energy,
www.doedigitalarchive.doe.gov/ImageDetailView.cfm?ImageID=1000801&page=search
&pageid=thumb, last update 1 June 1985, accessed 1 Aug. 2008.
[40]
"Free electron laser," Wikipedia, http://en.wikipedia.org/wiki/Free_electron_laser, last
update 2 June 2008, accessed 14 July 2008.
[41] P. Ramakoteswara Rao, "Laser isotope separation of uranium," Current Science, vol. 85,
no. 5 Oct. 2003.
40
FFI-rapport 2008/02376
Appendix A
A.1
Some relevant physics and chemistry
Isotopes
An atomic nucleus contains two types of particles: protons which have a positive electric charge,
and neutrons which have no electric charge. Around the nucleus there are electrons which have
the same charge as the protons only with opposite sign; they are negatively charged. In a neutral
atom there is therefore the same number of electrons around the nucleus as there are protons
inside the nucleus. It is mainly the number of electrons that determine the chemical properties of
an atom, and therefore different atoms which have the same number of protons (and therefore
also electrons) constitute the same chemical element. So when a chemical element is defined,
nothing is said about the number of neutrons. The number of neutrons in a nucleus is typically
about the same as the number of protons, but it may vary a little, and atoms of the same chemical
element, but with a different number of neutrons, are known as different isotopes of that element.
The main difference between two isotopes is their masses; the mass number of an atom indicates
the total number of protons and neutrons. When discussing isotopes, the mass number is generally
provided along with the chemical symbol to specify which isotope we are referring to. Uranium is
the heaviest naturally occurring element, and there are two important isotopes U-238 and U-235,
where U-238 contains three more neutrons than U-235. In naturally occurring uranium, U-238
amounts to approximately 99.3 % and U-235 only 0.7 %.
A.2
Nuclear fission and chain reaction
A nuclear power plant and a bomb are based on the same principle, the principle of fission of an
atomic nucleus. Some large nuclei are unstable and can sometimes split into two smaller nuclei of
roughly equal size while emitting a few independent neutrons; this process is called spontaneous
nuclear fission. These new nuclei and neutrons will have a total mass that is smaller than the mass
of the original nucleus. This mass defect is compensated by an energy release in accordance with
Einstein’s famous equation E = mc2 where c is the velocity of light (300 000 km/s). This is the
energy that is utilised in a power plant or a bomb. The uranium isotopes have a low probability of
spontaneous fission 21 (if spontaneous fissions occurred readily, it would not be possible to use
uranium in a reactor). When U-235 captures an extra neutron, it may fission and release two or
three new neutrons. If these new neutrons are captured by another nucleus and subsequently lead
to a new fission, a chain reaction is started. In a nuclear reactor such chain reactions are
controlled to give an approximately constant release of energy, whereas in a bomb the chain
reaction is allowed to run wild. For a chain reaction to be sustained, on average at least one of the
freed neutrons from each fission has to lead to a new fission. For this to occur there has to be a
certain amount of uranium present, and this minimum amount is known as the critical mass; if the
21
The partial half lives due to spontaneous fission for U-238 and U-235 are 9 x 1015 y and 1.2 x 1017 y,
respectively [37].
FFI-rapport 2008/02376
41
Figure A.1 Phase diagram of uranium hexafluoride (copied from
http://web.ead.anl.gov/uranium/guide/prodhand/sld022.cfm).
mass is greater it is called supercritical. Alternatively, it is possible to compress a smaller mass to
bring the atoms closer together and thereby reach a critical density. Since there is so little U-235
in natural uranium, it is necessary to enrich the uranium in this isotope for the material to be able
to sustain a chain reaction. Otherwise the critical mass or the critical density would be too high to
reach in practice.
A.3
Uranium hexafluoride
UF6, uranium hexafluoride, also referred to as “hex,” is a substance with good properties for
uranium isotope separation. Each molecule consists of only one uranium atom and six fluorine
atoms, and because fluorine only has one stable isotope, F-19, there are only two possible
compositions of the molecule, each with a specific molecular mass dependent of the uranium
isotope. The mass numbers of the molecules containing the different isotopes of uranium are 352
and 349, respectively, so the mass difference is three units which amount to only 0.85 %. Because
the mass difference comes directly from the different masses of the uranium isotopes, uranium
hexafluoride is a good choice for separation of these isotopes, but because the relative mass
difference between the molecules is so small, the enrichment will be less effective than if the
mass difference had been greater.
At room temperature and atmospheric pressure, uranium hexafluoride is a crystal, but at 56.5˚C it
turns directly into a gas without first becoming a liquid (it sublimes) (see the phase diagram in
Figure A.1). The molecule is organized with the uranium atom in the centre and the six fluorine
atoms spaced symmetrically around it, giving the molecule three perpendicular symmetry axes.
This makes it possible to make theoretical calculations of the vibrational energy states which are
utilised in the laser isotope separation methods. UF6 is highly reactive and reacts violently with
42
FFI-rapport 2008/02376
water and organic compounds. UF6 is also highly corrosive, and nickel or aluminium alloys are
needed to handle the gas. [2]
Uranium hexafluoride is used in gaseous form in centrifuges, in the gaseous diffusion method, in
the aerodynamic processes and in molecular laser isotope separation. It is used in liquid form in
the thermal diffusion process. To obtain the liquid form of uranium hexafluoride, the pressure
must be raised to just over atmospheric pressure, as can be seen from the phase diagram.
A.4
Statistical description of the molecules in a gas
A gas consists of a very large number of molecules that are moving around all the time at
different speeds and exchanging energy through collisions with each other. Because there are so
many molecules in a gas, it is impossible to describe the properties of the gas by considering each
molecule separately; instead it has been found fruitful to use a statistical description. This means
that while we cannot determine the exact speed or the exact energy of each molecule, we can
determine the average or the most probable values. The actual values for each molecule will
generally not be the same as this most probable value. Because of collisions between the
molecules, energy will be exchanged between them and the energy of individual molecules will
thus vary.
If the gas is in thermal equilibrium (in practice this means that the temperature and the pressure
are stable), all molecules have the same kinetic energy on average. The kinetic energy of a
molecule is given by E = ½mv2 where m is the mass of the molecule and v is its velocity. This
means that molecules with smaller mass must have a larger velocity to match the kinetic energy
of the heavier molecules. But this is only on average; the heavier molecules may have higher
velocities at some times, and the lighter molecules may have lower velocities.
A.5
Quantum mechanics
On the atomic and molecular level nature is not smooth. For instance a system (like an atom or a
molecule) cannot have a continuous energy spectrum, but can only take distinct, discrete energy
levels. Transitions between these discrete states occur with emission or absorption of a quantum
of energy; these transitions are called quantum leaps and are the smallest changes possible. This
quantum of energy carries the exact energy that separates the two states and is usually in the form
of a photon. Photons are the particles that light and other electromagnetic waves such as
microwaves, infra red light, visible light, ultraviolet light and x-rays consist of. A light ray is a
bundle of these particles, and each of them carry the energy E = hν where h is a scaling constant
and ν is the frequency of the light. This means that microwaves, x-rays, blue light and red light all
consist of photons with different energies. Electromagnetic waves with photons of high energy
have high frequencies (and correspondingly short wavelengths). By sending photons with the
exactly right energy (this means electromagnetic waves with the exactly right frequency) at an
atom or a molecule, it can be excited (go to a given state with a higher energy).
FFI-rapport 2008/02376
43
In an atom, the electrons around the nucleus occupy discrete energy levels. By adding energy to
an atom, an electron can be excited into a higher energy state. But for this to occur, the energy
which is added has to mach the energy difference between the old and the new energy levels.
Electrons can also be ejected completely if enough energy is added and this will happen easier if
the electron is already excited. When an electron is ejected the atom becomes an ion. The energy
that is added to ionize the atom does not have to be specific as long as it is large enough; spare
energy will just become kinetic energy for the electron (energy that gives it a velocity). An
ionized atom has a net electric charge which can be exploited if we want to influence its motion.
The energy levels of the electrons in an atom are determined by the nucleus of the atom.
Therefore there will be slight differences in the energy levels of different isotopes of the same
chemical element. By choosing photons of the exactly right energy, it is possible to excite and
ionize only one isotope in a mixture.
In molecules, the energy levels are mainly determined by the chemical bonds between the atoms.
When two or more atoms are bound together, they may vibrate relative to each other, leading to
discrete vibrational modes with distinct energies. When a molecule is excited, there is a change
from one vibrational mode to another, and the energy added to excite it has to mach the energy
gap between the modes exactly. If more energy is added to the system, the chemical bond can be
completely broken, and new chemical properties are created. If the energy is added through a
photon, this is called photo dissociation of the molecule. As mentioned above, the energy levels
of the vibrational modes of a molecule are determined by the atoms involved in the chemical
bond. There can be slight differences between the energy levels of a certain chemical bond
depending on the isotopes of the chemical elements involved. Therefore it is possible to excite
and dissociate only molecules containing a certain isotope if the photons can be tuned accurately
enough.
A.6
Oxidation states
The oxidation state gives the degree of oxidation of an atom in a chemical compound. It is equal
to the hypothetical charge that an atom would have if all bonds to other elements were completely
ionic (complete exchange of electrons). An increase in oxidation state is known as oxidation and
a decrease in oxidation state is called reduction. The algebraic sum of the oxidation numbers of
all the atoms in a compound is always equal to the net electric charge of the compound. The term
oxidation state is often used interchangeably with the terms oxidation number, even though the
actual definitions are slightly different, the values are usually the same. A given chemical
elements may have different oxidation states in different compounds; this is the origin of the term
redox reaction. A redox reaction is shorthand for reduction-oxidation reaction and is a chemical
reaction where the oxidation states of the atoms involved are changed.
Substances that are oxidative are known as oxidizing agents, oxidants or oxidizers, in a redox
reaction the oxidizing agent will be reduced. Substances that are reductive are known as reducing
agents, reductants or reducers, the reducing agent will be oxidized in a chemical reaction.
Uranium can be found in oxidation states +2, +3, +4, +5 and +6 [38].
44
FFI-rapport 2008/02376
Appendix B
Lasers
This appendix discusses in more detail the laser requirements necessary for laser isotope
separation.
In the atomic vapour laser isotope separation (AVLIS) technique (Section 8.1), to be able to
excite only one isotope, the lasers must be tuned with an accuracy in wavelength of one part in
100 000. This is possible with modern lasers. Especially a type of lasers called dye lasers, which
use organic dyes, may be used. High-power green or yellow light from copper vapour lasers (cf.
Figure 8.2) is fired into streams of special dyes, which convert the laser energy into beams tuned
to the exact colour absorbed by U-235 [39]. Four slightly different beams are needed (cf.
Figure 8.1); it is not possible to distinguish between these beams with the human eye. The light
that causes the desired transitions has a red-to-orange colour and can be made by use of a
common dye called rhodamine 6G. Because excited atoms will decay back down into a lower
state, the lasers must be intense enough to re-excite the ones that decay. The lowest acceptable
power density is on the order of tens of kilowatts per square centimetre. Because the uranium
atoms move through the radiation zone, the laser pulses must come often enough so that all the
atoms may be hit by them. The problem sufficiently high frequency pulses can be solved by using
several lasers that are synchronised such that the total frequency will be high enough. [26]
In the molecular laser isotope separation (MLIS) technique (Section 8.2), the required wavelength
is 16 µm, and there are several possible lasers. Powerful infrared lasers which use CO or CO2 are
available in many frequencies, but the infrared frequency that is needed to dissociate the uranium
hexafluoride molecule is harder to make [26].
The candidate which seemed to be the best alternative when the technique was developed, uses
Raman scattering from parahydrogen to increase the wavelength of the 10 µm light from carbon
dioxide. Raman scattering refers to an interaction of photons with molecules in which the
scattered photons have different energy than the incident photons; the change is determined by
the molecules’ rotational or vibrational energy levels. [28] Parahydrogen is a state of the
hydrogen molecule (H2) in which the two atoms have antiparallel spins.
Another option is to use a free electron laser. Free electron lasers are widely tuneable and used for
many purposes other than isotope separation, so there is a lot of activity in that area. In a free
electron laser, the electrons are accelerated to relativistic speeds and pass through a transverse
magnetic field giving them a sinusoidal path. This acceleration results in the release of photons.
Today, a free-electron laser requires the use of an electron accelerator with its associated
shielding, as accelerated electrons are a radiation hazard. The accelerator requires a high voltage
supply. Usually, the electron beam must be maintained in a vacuum which requires the use of
numerous pumps along the beam path. Free electron lasers can achieve very high peak powers.
Their tuneability makes them desirable in several disciplines. [40] Other types of lasers were also
suggested at the time the process was developed [28].
FFI-rapport 2008/02376
45
After the molecules have been excited by the infrared laser, an ultraviolet laser is used to
dissociate them completely. There is a larger probability for the excited molecules to be
dissociated by the ultraviolet photons than for the unexcited ones. The difference is not very
large, and this affects the efficiency of the method and implies that the ultraviolet laser must be
powerful [28].
A way to avoid the difficulties of finding a suitable laser is to use a molecule which fits with a
laser that is already developed. This has not been given much attention because of the preference
for uranium hexafluoride, but it has been proven possible. A successful demonstration has been
given [41] of uranium-isotope selective dissociation of a molecule with a CO2 laser, but the
molecule was very complex, and the work was ended because this approach was considered
unlikely to be competitive with approaches based on uranium hexafluoride or AVLIS.
46
FFI-rapport 2008/02376